Hermes Cloud: agents that work with the laptop turned off

What I wanted to fix

In the previous post, I showed how I installed Hermes Agent on my laptop and set up independent profiles to separate each assistant’s tasks.

Running agents directly on my computer gave me clean directory structure and total control over each agent’s rules. But I soon hit a basic physical barrier: my own daily schedule.

I work during the day, and at night and on Sundays I build with AI agents. My free time in front of the screen is limited. Whenever I closed my laptop to leave for work or went to sleep, every running task stopped immediately. If an agent needed to complete a long research run, organize files, or execute a routine at a fixed hour in the morning, my computer had to stay open on the desk, plugged into the wall, and connected to home Wi-Fi.

Keeping a laptop running all night was not a good idea. The machine ran warm, consumed power pointlessly, and any brief drop in home internet interrupted the work halfway through.

Between September 20 and 23, 2026, I decided to fix that problem. Instead of relying on my personal computer for everything, I turned on Hermes Cloud to keep assistants working quietly in the background all day long.

Quick summary

• In the previous post, I showed how I installed Hermes Agent on my laptop and created independent local profiles.
• Between September 20 and 23, 2026, I turned on Hermes Cloud, the version of the system that operates in the cloud 24 hours a day.
• Cloud agents keep working while you sleep, work, or commute, without requiring your laptop to stay powered on.
• I moved only scheduled tasks and routines needing continuous operation to the cloud, such as the newsroom team.
• The main lesson for beginners is to keep hands-on work on your local computer and move only self-running tasks to the cloud.

First: what the cloud and Hermes Cloud are

For anyone outside the technology field, the word cloud sounds abstract, but the reality is simple.

The cloud is simply someone else’s computer. Instead of sitting on your desk collecting dust, that computer lives inside a secure data center equipped with continuous power generators, proper cooling, and an internet connection that never drops. When you use a cloud service, you are simply renting a piece of that remote machine.

Hermes Cloud applies that same logic to AI agents. It is Hermes Agent running on a secure server on the internet, available all the time. While Hermes Desktop depends on keeping your laptop lid open, Hermes Cloud stays active even when every computer in your house is powered down.

A good comparison is the water supply in your home. Running an agent on your laptop is like drawing water from a hand pump in your yard: you have to walk over, pump the handle, and fill the bucket right when you need it. Having an agent in the cloud is like connecting your house to the municipal water main: pressure stays in the pipes all day, ready to flow whenever a faucet or an automatic garden timer turns on.

The comparison stops working at one key technical point: city water only flows when a valve opens on purpose. A cloud AI agent can be scheduled to act on its own. If its instructions lack strict stopping conditions, it will continue making system calls and processing data even if you forgot it was running.

What I used

Between September 20 and 23, 2026, I built this setup with straightforward tools:

• My personal laptop, where I kept Hermes Desktop for work I handle manually.
• Hermes Cloud, running in a remote environment for continuous execution.
• Task schedulers that trigger instructions on specific days and times.
• AI models configured to answer agent requests.

How I split the work

When you get access to a server, the temptation is to dump everything there without thinking. I avoided that by drawing clear boundaries.

On the laptop, I kept agents tied directly to my local machine:

1. Optimus: the leader of local agents on the laptop, receiving commands when I sit at the computer and managing local files.
2. Local Security: watches over the security of the laptop itself, reviewing permissions and files before direct execution.

On Hermes Cloud, I placed only agents that required autonomy and automated schedules:

• The newsroom: a team of agents with divided roles (such as gathering facts, checking, writing, art, and editing) to produce drafts methodically.
• Scheduled tasks: routines that need to run at set hours, such as checking information or organizing files early in the morning.

With that separation, my routine gained predictability. If a task needed to start at six in the morning while I was getting ready for work, Hermes Cloud ran the routine on the server, completed the planned steps, and had the results ready for review when I sat down.

What went wrong and what I learned

My initial instinct was to transfer every single profile from the laptop to the cloud all at once. Having everything in one central place seemed convenient.

In practice, that did not work. Placing an agent in the cloud for tasks dependent on local files meant constantly transferring folders back and forth over the internet. What was supposed to save time created unnecessary overhead.

Testing new ideas directly in the cloud was also slower. If an instruction file on the laptop has a typo, you open the folder, edit a single line in a text editor, and retest in ten seconds. In the cloud, you have to update the remote server, restart the service, and inspect logs in another window.

The lesson was direct: the cloud does not replace your personal computer. It extends it for tasks that need independence. Anything you are still experimenting with, designing, or operating by hand belongs on your local machine. Only move a workflow to the cloud once it is stable, proven, and thoroughly tested.

Costs and limits

Letting agents run on their own requires attention to practical limits.

On a personal laptop, the main cost is the electricity used and the model tokens consumed during an active conversation. In the cloud, continuous operation demands extra caution: if you configure an agent to run every ten minutes and it gets caught in a loop over poorly written instructions, it will burn through resources unchecked while you sleep.

There are also clear technical limits:

• A cloud agent has no visibility into what sits on your physical desk. It cannot see your local folders unless you set up an explicit file transfer.
• It depends entirely on provider network stability and external API availability.
• Cloud tasks require much stricter stopping rules, because no human is staring at the screen to hit cancel if something drifts off track.

What I would do differently

If I were starting this step today, I would wait longer before spinning up agents on a server.

I would refine every workflow thoroughly in Hermes Desktop first, running tasks manually for at least a full week. Making sure an agent knows exactly how to complete its steps from start to finish saves hours of troubleshooting strange behavior on a remote server.

I would also establish tighter circuit breakers from day one, ensuring no automated routine makes repeated attempts on failure without alerting me first.

Who this is for and who it is not for

Running agents on Hermes Cloud is useful for:

• People with limited free time during the day who need scheduled tasks to run reliably at set hours.
• Projects where a team of agents works through sequential handoffs without waiting for someone to open a laptop.
• Anyone who has already validated agent workflows locally and knows their instructions are clear and stable.

It is not meant for:

• Beginners who are still learning the basics and have never managed an agent manually.
• People who only want occasional back-and-forth chat in a browser to answer quick questions.
• Tasks that require continuous human supervision at every step or rely on local files that change constantly.

Wrapping up

Turning on Hermes Cloud between September 20 and 23, 2026, gave me the freedom to shut down my laptop without stalling progress on my projects. Knowing that scheduled tasks and the newsroom team were working steadily in the cloud changed the pace of my learning.

With agents running on both the laptop and the server, a practical need arose to set clear boundaries: defining who does what each day and which responsibilities must never leave human hands.

In the next post, I will share how our company meeting went: who does what, and what the human never gives up.

If you are building routines with AI agents or planning the split between local computers and the cloud, find me on X: @RodrigoesPorto.

Sources and links

• Official Hermes Agent documentation: https://hermes-agent.nousresearch.com/docs
• My official profile on X: https://x.com/RodrigoesPorto